The measurement of high speed electrical signals can be performed by sampling the signals at a series of time delays and then plotting signal amplitudes as a function of time. So-called “real time” digitizers typically have sampling rates no greater than about 1–2 Gsample/sec so that electrical signals having frequency components at frequencies greater than a few GHz must be characterized using so-called “equivalent-time” sampling. In equivalent-time sampling, a periodic input signal is sampled at a rate that is much less than the highest frequency component of the input signal over many repetitions of the input signal, and the measurements are assembled to provide an estimate of the input signal during a single period. Equivalent-time sampling is described in, for example, Marsland et al., U.S. Pat. No. 5,378,939 (“Marsland”) which is incorporated herein by reference.
For measurement of very high bandwidth electrical signals, equivalent-time sampling systems typically attempt to provide a short duration “strobe pulse” to one or more sampling diodes. The sampling diodes are switched by the strobe pulse, and then a portion (i.e., a sample) of the input signal is communicated to a signal acquisition system. The duration and magnitude of the sample is determined by one or more temporal properties of the strobe pulse, such as rise time, fall time, or duration. Accordingly, for high speed electrical signals, the strobe pulse should have a short rise time, fall time, or duration. Examples of sampling systems and strobe pulse generators for such sampling systems are described in, for example, Marsland, Rodwell et al., U.S. Pat. No. 5,014,108, McEwan, U.S. Pat. No. 6,060,915, Lockwood, U.S. Pat. No. 4,654,600, Lockwood, U.S. Pat. No. 3,760,283, Frye, U.S. Pat. No. 3,629,731, W. M. Grove, “Sampling for oscilloscopes and other RF Systems: Dc through X-band,” IEEE Trans. Microwave Theory and Technique MTT 14:629–635 (1966), and W. C. Whitely et al., “50 GHz sampler hybrid utilizing a small shockline and an internal SRD,” IEEE MTT-S Digest (1991), which are incorporated herein by reference.
While a fast strobe pulse is needed for such a sampling system, it is also desirable that the connection of an input signal to the sampling system neither introduce signal artifacts nor disturb the signal under test. Sampling systems establish a sample window by switching a sampling diode between conducting and non-conducting states with a fast strobe pulse, and typically a portion of the strobe pulse is transmitted to the device under test. This portion is referred to as “strobe kickout.” In addition, a portion of the signal to be measured is typically transmitted around one or more sampling diodes and detected even with the sampling gate closed. This signal portion is referred to as “blowby.” It will be apparent that signal artifacts caused by strobe kickout and blowby are preferably avoided. Other signal artifacts are caused by the connection of the sampling system to the signal to be measured. For example, the propagation of high speed electrical signals depends on the waveguide properties of cables and transmission lines on which the electrical signals propagate, and the connection of a sampling system to a cable or a waveguide generally loads the waveguide or presents an unmatched impedance. As a result, electrical signals arriving at the connection are partially reflected and these reflections can appear as artifacts in the measurement of the signal, or can be transmitted to the signal source, thereby changing the signal presented to the sampling system. In some prior art systems, signal artifacts are introduced by connection of the sampling system to a device to be tested so that measurements are corrupted by the connection.
In addition to the problems listed above, the temporal resolution of sampling systems can be limited by strobe pulse duration, strobe pulse rise or fall times, or difficulties in transmitting a strobe pulse to a sampling gate without degradation. Some sampling systems permit sampling only at relatively low sampling rates so that signal acquisition requires measurements over many signal periods. With such systems, because only a small fraction of a signal is measured, data acquisition is slow and random noise in measurements cannot be efficiently reduced by signal averaging.
In view of these and other shortcomings, improved sampling methods, sampling apparatus, as well as methods and apparatus for connecting signal sources to sampling systems are needed.